Davos 2025 | Salil Parekh highlights Infosys’s strategy in a rapidly evolving tech world

Davos 2025 | Salil Parekh highlights Infosys’s strategy in a rapidly evolving tech world
In the face of rapid technological advancements and intense competition, Infosys MD & CEO Salil Parekh emphasised the need for continuous innovation and adaptability.

Speaking about the growing buzz around Agentic AI, which dominated discussions at the World Economic Forum (WEF) in Davos, Switzerland, Parekh stressed the need for constant innovation to stay relevant in this environment.

Agentic AI is a type of artificial intelligence that acts independently, making decisions, taking actions, and adapting to changes without needing constant human guidance.

In an interview with CNBC-TV18, Parekh explained that Infosys focuses on training its employees to handle new and emerging technologies. Parekh noted that while there were many foundational models in the market last year, only a few have proven successful. Infosys’s goal, he said, is to work with these leading models while preparing its workforce for future changes in technology.

Below is the verbatim transcript of the interview.

Q: Being artificial intelligence-first (AI-First), we spoke about that a year ago. How do you back up that claim? Give me three numbers that back up the AI-First claim for Infosys.

A: There are two things that we are doing which we think are quite different. One, we are building four small language models. And these small language models are things that our clients are much more interested in today. These are models based on their proprietary data and our proprietary data. For example, we built one small language model for banking, and this is getting traction, where clients are starting to use this in their environment, another one for IT operations and so on. The second thing we are doing is building agents. And agents, today we are in the process of developing about 100 new agents.

To give an example, we built an agent for doing research work, and this is in production with a client for a technology company where their internal people and their external customers can research what the product specifications are and changes are by using an agent and huge benefit to them, things that used to take 14-15, days, now down to eight days and other such metrics. Another set of agents we built for an audit company. There we built three agents which are looking at their full audit process, and again, changes in there.

Also Read | Davos 2025: Infosys to hire over 20,000 freshers next fiscal; 2.7 lakh employees are now ‘AI aware’

So, our view of being AI-First, is first, putting it for clients and then inside. So, all our service offerings have been infused with AI. We built AI road maps. We are doing AI, which is secure by design and responsible AI where we want to make sure everything that we can build in, is built in to avoid what could be not good uses of AI. We are doing it at the start itself.

Q: How many people are now trained? The last time we spoke, you had given me a number. What does that number look like today, in terms of the people who have been trained on AI within Infosys?

A: Within Infosys, it is a huge number because there are two types of training. One is what we call AI awareness, and one is a deeper training where people are using it. The AI aware, which is a broad training, is over 270,000 out of our 320,000 people in this AI training, and it’s really across the board, because every discussion with the client, every deal, every large deal, every medium deal, there is a lot of interaction on AI. So, it’s built into many things.

Q: So clients are asking you now specifically for what the AI use case is going to be, and how you can contribute to that?

A: Clients are interested in where generative AI can help them. What we are seeing is through agents or small language models; in the past, through large language models, we have various ways in any large transaction or programme where they can benefit from this AI. And so, we are using it, and they are quite happy with it.

Q: What is this going to mean in terms of incremental spending by customers and by clients, and what will it translate into as far as revenue for you is concerned, as well? As we look at people moving from pilots now to scale and scale and speed, which is, of course, the theme here in Davos when we talk about AI adoption, what will it mean in terms of a bump up, both in terms of revenue as well as spending?

A: There are two types of generative AI. One is the consumer, where we see all this discussion here also the World Economic Forum, but everywhere, the massive capex that we are hearing about. Our focus is more on this enterprise generative AI, which is a smaller amount because the consumer one is huge.

Also Read | Davos 2025: TCS chief Krithivasan says AI may create more jobs, not take them

In enterprises, we are seeing the traction is large, but the revenue size is still not that large. It is quite small, but it is increasing. So, every quarter we see good traction and progress on what is going on. My sense is, if you sort of look back to when cloud started or when digital started, it starts with enterprises, where there is a lot of interest, and then there is an inflexion point, and then cloud picks up, or digital picks up.

Q: When do you expect the inflexion point for generative AI at the enterprise level?

A: Difficult to say when that time that inflexion point will come. But the way it normally works is you keep building different areas where impact is created, and then suddenly there comes a time when it becomes quite overwhelming, and the inflexion point comes about.

Q: So, are we 12-18-24 months away from possibly that inflexion point?

A: Difficult to say when that time will come. But if you look at digital as an example. It started maybe eight to nine years ago. Today at Infosys, over 60% of our work is all digital, and so it’s a big change from where we were, let’s say 5-10% but it is difficult to say when exactly that time will happen.

Q: What does this now mean in terms of workforce, I have been asking your peers this question, and most seem to suggest that it is not going to mean a smaller number of people being hired. It’s probably going to be the same, if not more. What is your assessment? Are we likely to see a headcount reduction? Are we likely to see a status quo? Are we likely to see you hire more to be able to deliver on the capabilities you spoke of?

A: In the October-December 2024 quarter (Q3FY25), we had a 5,000-person increase in headcount, 5,600. For the coming financial year (FY26), we have estimated in the range of 20,000 college hires. So, all of this is pointing to an expansion.

So why the expansion? Part of what happens with generative AI is you get more productivity benefits, but a lot of what is happening is you have much larger areas of the enterprise where you can do work in generative AI. So, in that larger area, with the productivity, the net is still an increase in the headcount, and that is the trend we are seeing right now.

Q: So, you will continue to see an increase?

A: For this, we already have an early plan of about 20,000 college graduates.

Q: When you hear what is happening in the US now, when you look at the possibility of deregulation, which has been promised for sectors like BFSI, what will that mean? You have upped your guidance already. But do you believe that there is a potential for a further upward revision on the back of the changes that are likely to come in?

A: Absolutely. To hear about $500 billion, it is staggering what type of thinking they have in the US for technology and infrastructure, and this is all AI-driven. You have also seen lots of other capex announcements over time. We see a lot of discussion on regulations coming down.

The sense we get from people we talk to, and even here (at the World Economic Forum), the general view is the economy in the US will do better. We have seen in the past when the US or Western European economy starts to do better, our business starts to grow. So, we have two things, one is where we do more cost and efficiency, and the other is a lot of growth and digital and so that will start to play out more.

My sense is, if that US economic growth and so on happens, which is what it is now looking like. We will see more and more growth overall for Infosys as well.

Q: So possible upward revision of the guidance that you have already revised higher?

A: We only go quarter by quarter. We will have the end of the year, so there will be next year. But we did increase our guidance for this year because of what are the signs financial services were good – US and Europe, and then retail in the US started to look better than the previous quarter. So, the thing with the US as the economy, all these things filtered into the economy, we will see how it plays out.

Q: But specifically, as far as the BFSI side is concerned I am picking up a lot of optimism from the CEOs that we speak to. Brian Moynihan of Bank of America was talking about the possibility of deregulation. What are you hearing from them now on what they will perhaps need?

A: Their results are just incredible for some of the large US banks, even the European banks’ results we have seen over the last couple of quarters improving. My sense, at least what we hear from our clients in financial services, they will continue to focus on cost and efficiency, which they want to do, but with this, there is much more investment in new technology which can help them drive growth. So, to drive growth, for example, we could consider the building as an example, an agent that can help them analyze credit, that can help them analyze customer segmentation. So, there is so much new generative AI work that can happen, and as they start to spend more, which will help them drive growth, we will hopefully support them.

Q: One of the other interesting conversations that I have been having with some of your peers seems to suggest that perhaps Indian IT services also find themselves at an inflexion point today as the world is moving into a generative AI-driven world. In that context, the advantage that Indian IT services have enjoyed is being able to deliver efficiency at scale. But now will you need to be much more IP-led and IP-driven, and what will that mean as far as changing the DNA of your organisation is concerned?

A: No question is going to be much more focused on whether we call it IP or innovation. So, there is, with the way generative AI is moving, let’s say, to build a small language model, we have put together an innovative approach. It is an approach which takes existing technology and builds it on existing platforms and a much lower cost than a large language model. So that innovation, innovation in agents, innovation and where it can be applied will continue to increase. And that is something we have been doing at Infosys for some time. My sense is that is a good sign. Now more of that will happen as we go through the next phase, and we will continue to have, let’s say, the intense focus on cost and efficiency. So, both of those will go well.

Q: But will you need to speed up on the innovation side? And what will that mean in terms of the capital required, the capital deployment that will be needed to push this agenda forward?

A: For us, we are still very much a light asset or light business, in that sense. So, our investment, even the small language model we built, is significantly smaller, the investment, than what we have read and heard about what large language models take or other things. So, our focus is going to be more on making those innovations, but without doing something that will stretch us on our balance sheet. We have a good balance sheet, but we are not in a position to make these massive capital investments, and we will work within that balance sheet to make that innovation happen.

Q: Will that be enough? And I understand the constraints as far as the balance sheet is concerned but will that be enough is the question given the fact that we are seeing such rapid moves in technology and such a fiercely competitive space. You are talking about agents. Everyone here (at the World Economic Forum) is pitching the same story. Agentic AI is the buzzword, everyone is talking about it and doing it.

A: Absolutely. I think we have to find ways to innovate constantly in that environment. So, the thing that we over the years at Infosys have done well is make sure we have a trained workforce that can work on different technologies, typically with a lot of these other technologies, some will work well, and some will not.

Already today, in the foundation model world, we coalesce to maybe three or four that are doing well. Last year, when we spoke, there were at least 15-20 in the mix, and all of them have spent massive amounts. So, our objective is to work with all of them and make sure that our employees are skilled enough to work on the new ones that come up.

For the entire interview, watch the accompanying video

Jannik Sinner Aiming Join Roger Federer, Novak Djokovic In Elite Australian Open List Previous post Jannik Sinner Aiming Join Roger Federer, Novak Djokovic In Elite Australian Open List
Home ownership remains a big part of the American Dream Next post Home ownership remains a big part of the American Dream

Leave a Reply

Your email address will not be published. Required fields are marked *